Source: TechCrunch
Twitch is converting its archive of millions of hours of streamed video into training data for Amazon's generative AI—a legally permissible but ethically fraught move that treats creator output as raw material without prior consent. The opt-out structure means Amazon captures value from creators' labor by default, mirroring how platforms have historically extracted data while paying creators minimally or not at all. As AI becomes valuable, the question of who owns the training signal and whether creators should be compensated for it will intensify across video, text, and music platforms.